What does a repertoire even look like?

I was really surprised when eight days ago there was a burst of interest in repertoires, but not one person who posted their repertoire in those threads had a tree-like set of moves, but rather only a general description with opening names. Was this due to the (very justifiable) reluctance to disclose one's opening surprises for fear of the competition could prepare in advance? Or was it due to people changing their moves within certain variations? Or do people here not keep there repertoires in a detailed, tree-like format in a computer file?

https://www.chess.com/forum/view/chess-openings/your-repertoire
https://www.chess.com/forum/view/chess-openings/what-is-your-repetoire

I thought I'd show the format in which I keep my repertoire since it could be useful to a lot of people. I've posted only a theoretical repertoire here, not necessarily my own, so as to not disclose any possible opening surprises I might use. This format has worked well for me for that past 1-2 months, which is when I began this particular format. Before that I had a separate file for each extra depth of 1-2 moves, which resulted in a lot of files and I couldn't see everything at a glance. Like Fischer, I always use the same response every time, except in uncommon cases when I've given up on a given variation out of frustration, and/or when I'm in the process of making repertoire changes.

First, my repertoire is divided into two big sections: first as White, then as Black, all in the same text file. Also, I have a lot of embedded notes within braces {} in my own standard format with my own notation. I chose braces because that is compatible PGN so that I can cut and paste sections of my repertoire directly into PGN players, such as on this site, without having to edit anything first. As for my notes, those are a *huge* topic that I'll probably post some day in a different thread. Basically I'm trying to create an entirely new language for chess, a language that represents plans as well as moves, in a compact and consistent format, using what looks like Backus-Naur notation with math functions.

The great thing about always using the same set of opening responses in every game is that it *greatly* reduces how much you have to remember, plus greatly simplifies the resulting file in size and complexity. In this way a repertoire differs from an opening book, since an opening book should show all the main possibilities for both sides. For example, an opening book for Petrov's Defense would need be organized roughly as follows, where I've limited the number of responses at each step to a mere three:

If we took only the single most popular response for Black as Black's preferred move--his repertoire move--the above move tree for Black's repertoire would simplify to:

Quite a bit of space (and memory) savings! Note that there are two ply per printed line instead of just one ply, which is possible because the second choice on each line is fixed in such a repertoire.

As mentioned, my repertoire also includes various notes, all in braces. The example below shows the opening names in the way I put them:

If this repertoire were for White instead of Black, the only significant difference would be that each line would begin with Black's move instead of White's, such as:

An important practical consideration is a convention for when to stop including all the possible responses. For example, using the immediately above sample repertoire, 365chess actually shows all the following responses at the expanded branch at the bottom:

To list all those 17 historically recorded responses is a bit much. You can make your own convention to decide the cutoff point. The convention I'm currently using is: Don't include any games that are at or below 1% in popularity, *unless* I've actually encountered it before. To implement this convention, however, that mean time-consuming use of a calculator.

For example, the sum of the number of recorded games in that last branch is 389, so in my earlier version of my repertoire I would write, accurate to the nearest full percent...

Since the cutoff point is 1%, I would not include any moves from 4...Ng5 on down.

Yes, this is a lot of work, but I've kept such repertoires since the '80s, and I'm still using much of my old '80s repertoire, and I'm getting tired of writing it up from scratch every time my hard disk crashes, so I suggest people make backups and regard this as a very long-term project.

Also, in my newer, more practical repertoire, I don't use such notes about the percentages anymore, since I use percentage mostly only as a guideline when I'm trying to find my favorite response, and after I've decided on my favorite response, I don't care about its popularity. However, I still use other percentage notes that look something like:

12. c4 {n 36%.} dxc4 {nb 38%.}

In my convention, "n" means "nobody wins," so "n 36%" means nobody wins 36% of the time with that move, in other words a draw. "nb" means "nobody and Black" win the same percentage of the time. I do find those statistics useful because I can see at a glance the expected outcome and character of that line. I'm choosing only the most common outcome, rather than database-like "White wins x%, nobody wins y%, Black wins z%," since I care mostly about the most common outcome. As I said, there is *much* more I'm starting to notate in my cryptic fashion, but that is a huge topic.

Anyway, that's pretty much how *my* repertoires look, it seems to me that's a necessary organization, but maybe other people structure their repertoires differently.

You are right with your tree, but i would not waste my time to post such a tree here. I use some books, where i could copy the index, but that's not legal. Even worse would be to publish a pgn file with all exact lines.

A repetoire is just a shortcut attempt to winnow out all lines needed to play in a tournament without having to know everything. Eventually, you run into a situation where you didn't consider a possibility of transposition and suddenly you are in someone else's repetoire without a map.

Clearly you are unprepared!

Anybody that properly prepares their repertoire would take these transpositions into consideration.

FOR EXAMPLE:

My Repertoire as Black vs 1.e4 is the Taimanov Sicilian and the Najdorf Sicilian, so I could respond either way to what follows, but let's assume you are one that only plays the Najdorf.

What do you do after 1.e4 c5 2.Nc3?

2...d6 is dubious because it makes the Grand Prix attack better than its reputation (Black wants to play ...d5)

2...Nc6 is ill-advised because after 3.Nf3! (oops, guess White wasn't playing the Closed or Grand Prix Attack anyway!) and now White is getting ready to play d4, and you've put the Knight on the wrong square for your repertoire!

The answer is 2...a6!  If White stays with the Closed or Grand Prix, you have something like 3.g3 b5 4.Bg2 Bb7 5.d3 e6 6.f4 d5 7.exd5 b4 8.Ne2 exd5 and this line actually has a good reputation for Black.

Najdorf Players and Kan Players, along with O'Kelly Players of course, should answer 2.Nc3 with 2...a6.

If you just blindly consider 2.Nf3 and 2.Nc3 completely independently, then yes, of course you are going to have problems with move order tricks!  If you have enough sense to realize that 2.Nf3 and 2.Nc3 are not totally 100% independent of one another, and look at the big picture rather than treating everything you play like they are 100 rats in 100 separate compartments where no 2 can interact, then you wouldn't have that problem!

Some other common ones you must take into consideration:

1.e4 e6 2.d4 c5.  Now 3.Nf3 is the Open Sicilian, 3.d5 is the Franco Benoni.  You need to know one or the other, despite 2...c5 not being common as it's not busted either!

1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.f3 Nc6 6.c4 g6 7.Nc3 Bg7 is a direct transposition to the Sicilian Accelerated Dragon, a line you don't normally see with 2...d6.  Black has other independent options, but playing in pure Najdorf style with 5...a6 (a move I see often), is just bad for Black after 6.c4!

There are many others as well!

In my opening study, which I consider thorough, I usually do not include moves that have been played less than 5 times in my livebook.

I keep my repertoire in the excellent software Chess Position Trainer. It basically saves every move you put in and allows for nice overviews to be made.

Transpositions. Yes, another complication I didn't mention in my post, but I do have a way to handle those intelligently in my actual repertoire.

For example, if my repertoire for Black contained both Petrov's Defense and the Vienna Game (as it should), there would be a note in one of those openings to refer to the other, such as:

Note that one variation of the Vienna Game transposes to the Petrov Three Knights Game. "q.v." is Latin for "which see," which is a traditional way to direct the reader to a different part of the same document, equivalent to a clickable link.

Since repertoires take up a lot less text space than opening books, a thought occurred to me: I wonder if a trend will ever occur where respected master players will start publishing their repertoires, say after they retired, rather than opening books. Although such a repertoire would have the additional drawback that the reader would be "forced" to use only somebody else's move choice, if a repertoire were explicitly tailored to a given philosophy, say "draw as Black" and "avoid tactics if possible," then it's likely there would be often only one applicable move at each step anyway.

i have it all in my head. otherwise I just play some opening. I don't care to make such things..

There's a huge disadvantage with that approach: you will likely forget. Remember, most of us continue to play chess for the rest of our lives, repertoires for a good player are likely *huge*, and if you ever go through a period of several years where you lost interest in chess (this has happened to me multiple times in my life already!), you will certainly forget many of the moves you used to play.

When I took a course in the Pascal programming language, our instructor assured us that if we don't document our programs with comments, we *will* eventually forget what we were trying to do, and then we'll regret it. He was so right, as I found out year after year. That is extremely good wisdom--I've seen so many software companies that didn't enforce the practice of documenting, and all suffered very bad consequences, including at least one of them going out of business.

May I say it sounds paradoxical to me ?

I guess you have no problem with moves that are played quite often.
OTH, I personnally would write down some unfrequent moves which made me loose the game or spend a lot of clock time.

As someone who posted in at least one of those threads, I felt I should comment. I do sometimes keep a repertoire similar to yours in PGN files (one file per variation), but wouldn't want to share it here. I will however show one line from my 1. e4 repertoire because it's not a very good line for Black:

This is MUCH smallar than several other lines such as the Sicilian, 1...e5, the French, and the Caro-Kann, but it's similar.

Currently I don't have most of my current repertoire (the English, Slav, and Caro-Kann) in writing because:

1. It's easier to just not do it (bad reason).
2. During the last year I've been switching openings a bit.

As for the second point, in a way I think switching openings sometimes is good because you learn new patterns and about chess in general, but I also think it's good to try to really master an opening. At the moment I'm planning to stick to the Caro-Kann and Slav as Black and the English as White, but maybe play something different as White against lower rated opponents. In any case, I'm glad I have the PGN file with the 1. e4 repertoire, and should probably eventually make one for my current openings.

That brings up another point or two...

Another use of repertoires is for serious exploration. For example, since you mention the Caro-Kann, for months I had a lot of trouble finding lines I like to play as White in that opening, especially at the 7th move in one of the variations. Usually I try to select the most popular line, unless there's a line that draws much more easily, but I tried option after option for months and disliked all the resulting positions until I got down to the 7th most popular move. As soon as I experimented with it a little I knew that was the line for me. That happened in two other openings, too, where I finally selected the 8th most popular move, which is very rare for me: usually I end up choosing one of the first three most popular moves. It is during that period of investigation that you will want to be documenting the lines you played and why they didn't work, which is best represented in tree format.

Another point: During a period of such exploration I highly recommend reverting back to the one-ply-per-line-of-text format that was the first style I showed, otherwise you'll have to keep indenting as well as deleting and rewriting as you discover numerous faulty lines along the way, whereas if you just keep the one-ply-per-line method you never have to edit or delete any of the old lines, other than putting a comment like {Don't use this line.}.

One more related comment: Another convention I use in my repertoire comments is {pt.}, which stands for "pro tem," which is Latin for "for the time being." Whenever I see that comment, I know it's a flag to remind me to look for better moves for myself at that point because I haven't explored other options yet at that point. Note that I'm using highly abbreviated notation because I'm trying to squeeze as much useful information onto each line of text as possible. All my abbreviation keywords are only 1-4 letters long. The "plan language" I'm developing uses the same kind of notation, such as {th(wom).} to mean "threatens win of material." I'm reluctant to post my conventions on that, however, not because it's secret but because I started it only about one week ago, so it's going through rapid changes and probably contains highly inefficient oversights at the moment.

As a "former" Petroff player, I can easily tell you that you have already left out a major component that you must take into account, and I faced it twice in the US Open in 2014 (winning both).

The Exchange French!

After 1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nf3 Nxe4 5.d3 (instead of 5.d4 or 5.Nc3) 5...Nf6 6.d4 and Black has nothing better than 6...d5, which directly transposes to 1.e4 e6 2.d4 d5 3.exd5 exd5 4.Nf3 Nf6, just with 2 extra moves each.

All the sample repertoires shown here are merely theoretical, limited to exactly three responses for demonstration purposes only, and are not necessarily my own choices of moves. I am very well aware of the French attack of the Petrov, for example, but that was not in the top three responses in the database that I was using as an example.

Just play by first principles. Don't follow books, get your king castled and formulate a plan and you will be fine.

No, I will *not* be fine with first principles. I lose all the time to Computer Level 10 on this site, largely because I don't know the openings or the more subtle positional details such as what to do with specific pawn structures, or where to place my pieces, or what the general plans are. That's why I started to create my own plan language: to ease that transition from opening to middlegame. That's a much deeper topic that I'll probably post in another thread, though.

This is an interesting thread and I am happy that some have shared their findings, even if they are just examples.

As a (probably at best) average "hobby" player, I spend minimum time on theory and tend to play the same lines that are in my head for a given opening/opening response. I have only recently started paying attention to the minefield of the Sicilian (from White's perspective - I never choose it as Black because of my ignorance!) and have had some surprising success. On this site my stats are probably about 50% against similarly rated players from bullet to online for all games. I will openly admit that I prefer Blitz play just for fun and winning (or losing) is merely a by-product of me enjoying the game.

What I take from this thread is that, for a relatively small investment of time, I can literally map out moves through the opening to give myself a playable position that I actually understand. I am sure that I have accidentally reached playable positions in unfamiliar openings before but then stumbled on what to do next.

Thanks for the direction!

Well it's up to you how far you want to go. Deeper on in the opening I will be more likely to include rarer options, especially in the KID and the Dragon, but in the early stages when a move has been played less than 5 times I know it should be sub-optimal at best. Then I can simply use my general knowledge of the opening to get a good position.

It is impossible for me to forget my opening reportarie, as I do not have certain lines. I learn the IDEAS of the openings I play, and then I figure out if there are any traps in my move order, if not then I am happy to play my ideas and not certain first 10 moves without any idea why I played these moves.

Like here I took on d5, with the intention of getting strong pawns in the center.

My definition of the repertoire is the knowledge, tactics, and strategies an individual player brings to the table. As for where I store it, on the data base that I have with me at all times, my mind.